Building Intelligent AI Agents Powered by Claude to Automate Complex Workflows
SXMN is architecting the next generation of collaborative AI agents, designed to simplify operational friction, execute deterministic tasks, and empower modern enterprise teams.
Our AI Agent Ecosystem
A modular network of purpose-built AI agents coordinated by an intelligent orchestrator to execute end-to-end enterprise tasks with verified accuracy.
Development Status Note: All agent modules listed below are currently in the research and architectural design phase ("Planned"). We do not claim active commercial availability or live customer deployment.
Research Agent
Deep Knowledge Synthesis & Citation VerificationConducts multi-source inquiries across internal repositories and authorized web benchmarks, synthesizing domain research with strict citation traceability.
Workflow Automation Agent
Cross-System Process Execution & Event RoutingConnects business communication platforms, ERP backends, and databases using deterministic state machines and event-driven trigger pipelines.
Document Intelligence Agent
Multi-Format Parsing & Structured Schema ExtractionIngests unstructured PDFs, scanned invoices, receipts, and spreadsheets, outputting validated JSON records with confidence scores.
Customer Support Agent
Contextual Triage & Diagnostic EscalationAnalyzes inbound customer requests, checks knowledge repositories for verified answers, and drafts contextual summaries for support engineers.
Data Analysis Agent
Tabular Synthesis & Risk Metric CalculationProcesses business transaction tables, calculates variance benchmarks, flags anomalies, and generates executive summaries.
Multi-Agent Orchestrator
Task Graph Planning & Dynamic ConsensusDeconstructs complex high-level goals into directed acyclic task graphs (DAG), assigns subtasks across domain agents, and verifies convergence.
Native Adaptive Thinking active across 1,000,000 token context window for DAG task decomposition.
End-to-End Multi-Agent Architecture
Explore how user requests traverse from natural language intake, through Claude orchestration, to specialized agents and deterministic execution.
2. AI Orchestrator (Claude): Task decomposition & execution graph planning
Claude models analyze context, break goals into parallel or sequential subtasks, and dynamically route assignments to specialized agents.
How Claude API Powers Our Agent Architecture
We are evaluating Anthropic Claude as the foundational reasoning and coordination engine for complex enterprise automation.
Advanced Reasoning & Planning
Claude excels at interpreting ambiguous business instructions, reasoning through multi-step edge cases, and constructing verifiable execution graphs without hallucinating facts.
- Systematic hypothesis testing and task decomposition
- Superior performance on complex coding and domain reasoning benchmarks
- High adherence to nuanced system directives and organizational policy constraints
Explore Multi-Agent Collaboration & Live Claude API
Test task decomposition, RAG citations, and anomaly scoring either via instant architectural benchmarks or real-time Anthropic Claude API calls.
Flexible Dual Execution Modes: Supports dual modes: Instant Architecture Benchmark (runs offline without credentials), or Live Claude API calls (Anthropic Messages API endpoint) with client key or server ANTHROPIC_API_KEY. Never logs customer secrets.
Default: Automatically runs verified architectural simulation for zero API friction, or switches to live Claude calls when a key is provided.
// Click "Execute Agent Pipeline" to dispatch request and track agent handoffs...
From Conception to Intelligent Automation
Started in June 2022 in Vietnam, SXMN began as an initiative to tackle operational friction in repetitive business tasks. Observing how modern teams spend countless hours on manual document sorting, data extraction, and cross-application coordination, we set out to explore pragmatic software solutions.
Over the course of our development journey, SXMN has focused on evaluating modern language models and autonomous agent workflows. Our priority remains delivering transparent, reliable, and privacy-conscious tools tailored for enterprise productivity.
Project Facts
What We're Building
Three exploratory product tracks in active design and evaluation.
AI Workflow Automation
AI-assisted workflows for repetitive business operations and team coordination.
- Cross-app data reconciliation
- Triggered operational task flows
- Error detection & summaries
Intelligent Document Processing
Tools for extracting, indexing, and organizing structured information from business documents.
- Contract and invoice data extraction
- Multi-format file classification
- Audit trail verification
AI Business Assistants
Context-aware assistants designed to help teams retrieve knowledge and complete tasks securely.
- Internal policy inquiry
- Customer support triage assistance
- Structured report generation
AI at the Core of Our Architecture
We design systems using structured output validation, retrieval-augmented generation (RAG), and deterministic safety layers.
Transparent Development Roadmap
We believe in radical clarity. Track our progression across research, architecture evaluation, and future MVP releases.